| Message ID | 20230302235356.3148279-12-robdclark@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | dma-fence: Deadline awareness | expand |
On Thu, Mar 02, 2023 at 03:53:33PM -0800, Rob Clark wrote: > From: Rob Clark <robdclark@chromium.org> > > For an atomic commit updating a single CRTC (ie. a pageflip) calculate > the next vblank time, and inform the fence(s) of that deadline. > > v2: Comment typo fix (danvet) > > Signed-off-by: Rob Clark <robdclark@chromium.org> > Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> > Signed-off-by: Rob Clark <robdclark@chromium.org> > --- > drivers/gpu/drm/drm_atomic_helper.c | 36 +++++++++++++++++++++++++++++ > 1 file changed, 36 insertions(+) > > diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c > index d579fd8f7cb8..d8ee98ce2fc5 100644 > --- a/drivers/gpu/drm/drm_atomic_helper.c > +++ b/drivers/gpu/drm/drm_atomic_helper.c > @@ -1511,6 +1511,40 @@ void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev, > } > EXPORT_SYMBOL(drm_atomic_helper_commit_modeset_enables); > > +/* > + * For atomic updates which touch just a single CRTC, calculate the time of the > + * next vblank, and inform all the fences of the deadline. > + */ > +static void set_fence_deadline(struct drm_device *dev, > + struct drm_atomic_state *state) > +{ > + struct drm_crtc *crtc, *wait_crtc = NULL; > + struct drm_crtc_state *new_crtc_state; > + struct drm_plane *plane; > + struct drm_plane_state *new_plane_state; > + ktime_t vbltime; > + int i; > + > + for_each_new_crtc_in_state (state, crtc, new_crtc_state, i) { > + if (wait_crtc) > + return; > + wait_crtc = crtc; > + } > + > + /* If no CRTCs updated, then nothing to do: */ > + if (!wait_crtc) > + return; Is there an actual point in limiting this to single crtc updates? That immediately excludes tiled displays/etc. Handling an arbitrary number of crtcs shouldn't really be a lot more complicated should it? > + > + if (drm_crtc_next_vblank_start(wait_crtc, &vbltime)) > + return; > + > + for_each_new_plane_in_state (state, plane, new_plane_state, i) { > + if (!new_plane_state->fence) > + continue; > + dma_fence_set_deadline(new_plane_state->fence, vbltime); > + } > +} > + > /** > * drm_atomic_helper_wait_for_fences - wait for fences stashed in plane state > * @dev: DRM device > @@ -1540,6 +1574,8 @@ int drm_atomic_helper_wait_for_fences(struct drm_device *dev, > struct drm_plane_state *new_plane_state; > int i, ret; > > + set_fence_deadline(dev, state); > + > for_each_new_plane_in_state(state, plane, new_plane_state, i) { > if (!new_plane_state->fence) > continue; > -- > 2.39.1
On Fri, Mar 3, 2023 at 7:12 AM Ville Syrjälä <ville.syrjala@linux.intel.com> wrote: > > On Thu, Mar 02, 2023 at 03:53:33PM -0800, Rob Clark wrote: > > From: Rob Clark <robdclark@chromium.org> > > > > For an atomic commit updating a single CRTC (ie. a pageflip) calculate > > the next vblank time, and inform the fence(s) of that deadline. > > > > v2: Comment typo fix (danvet) > > > > Signed-off-by: Rob Clark <robdclark@chromium.org> > > Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> > > Signed-off-by: Rob Clark <robdclark@chromium.org> > > --- > > drivers/gpu/drm/drm_atomic_helper.c | 36 +++++++++++++++++++++++++++++ > > 1 file changed, 36 insertions(+) > > > > diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c > > index d579fd8f7cb8..d8ee98ce2fc5 100644 > > --- a/drivers/gpu/drm/drm_atomic_helper.c > > +++ b/drivers/gpu/drm/drm_atomic_helper.c > > @@ -1511,6 +1511,40 @@ void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev, > > } > > EXPORT_SYMBOL(drm_atomic_helper_commit_modeset_enables); > > > > +/* > > + * For atomic updates which touch just a single CRTC, calculate the time of the > > + * next vblank, and inform all the fences of the deadline. > > + */ > > +static void set_fence_deadline(struct drm_device *dev, > > + struct drm_atomic_state *state) > > +{ > > + struct drm_crtc *crtc, *wait_crtc = NULL; > > + struct drm_crtc_state *new_crtc_state; > > + struct drm_plane *plane; > > + struct drm_plane_state *new_plane_state; > > + ktime_t vbltime; > > + int i; > > + > > + for_each_new_crtc_in_state (state, crtc, new_crtc_state, i) { > > + if (wait_crtc) > > + return; > > + wait_crtc = crtc; > > + } > > + > > + /* If no CRTCs updated, then nothing to do: */ > > + if (!wait_crtc) > > + return; > > Is there an actual point in limiting this to single crtc updates? > That immediately excludes tiled displays/etc. > > Handling an arbitrary number of crtcs shouldn't really be a lot > more complicated should it? I guess I could find the soonest upcoming vblank of all the CRTCs and use that as the deadline? BR, -R > > + > > + if (drm_crtc_next_vblank_start(wait_crtc, &vbltime)) > > + return; > > + > > + for_each_new_plane_in_state (state, plane, new_plane_state, i) { > > + if (!new_plane_state->fence) > > + continue; > > + dma_fence_set_deadline(new_plane_state->fence, vbltime); > > + } > > +} > > + > > /** > > * drm_atomic_helper_wait_for_fences - wait for fences stashed in plane state > > * @dev: DRM device > > @@ -1540,6 +1574,8 @@ int drm_atomic_helper_wait_for_fences(struct drm_device *dev, > > struct drm_plane_state *new_plane_state; > > int i, ret; > > > > + set_fence_deadline(dev, state); > > + > > for_each_new_plane_in_state(state, plane, new_plane_state, i) { > > if (!new_plane_state->fence) > > continue; > > -- > > 2.39.1 > > -- > Ville Syrjälä > Intel
On Fri, Mar 03, 2023 at 07:45:05AM -0800, Rob Clark wrote: > On Fri, Mar 3, 2023 at 7:12 AM Ville Syrjälä > <ville.syrjala@linux.intel.com> wrote: > > > > On Thu, Mar 02, 2023 at 03:53:33PM -0800, Rob Clark wrote: > > > From: Rob Clark <robdclark@chromium.org> > > > > > > For an atomic commit updating a single CRTC (ie. a pageflip) calculate > > > the next vblank time, and inform the fence(s) of that deadline. > > > > > > v2: Comment typo fix (danvet) > > > > > > Signed-off-by: Rob Clark <robdclark@chromium.org> > > > Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> > > > Signed-off-by: Rob Clark <robdclark@chromium.org> > > > --- > > > drivers/gpu/drm/drm_atomic_helper.c | 36 +++++++++++++++++++++++++++++ > > > 1 file changed, 36 insertions(+) > > > > > > diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c > > > index d579fd8f7cb8..d8ee98ce2fc5 100644 > > > --- a/drivers/gpu/drm/drm_atomic_helper.c > > > +++ b/drivers/gpu/drm/drm_atomic_helper.c > > > @@ -1511,6 +1511,40 @@ void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev, > > > } > > > EXPORT_SYMBOL(drm_atomic_helper_commit_modeset_enables); > > > > > > +/* > > > + * For atomic updates which touch just a single CRTC, calculate the time of the > > > + * next vblank, and inform all the fences of the deadline. > > > + */ > > > +static void set_fence_deadline(struct drm_device *dev, > > > + struct drm_atomic_state *state) > > > +{ > > > + struct drm_crtc *crtc, *wait_crtc = NULL; > > > + struct drm_crtc_state *new_crtc_state; > > > + struct drm_plane *plane; > > > + struct drm_plane_state *new_plane_state; > > > + ktime_t vbltime; > > > + int i; > > > + > > > + for_each_new_crtc_in_state (state, crtc, new_crtc_state, i) { > > > + if (wait_crtc) > > > + return; > > > + wait_crtc = crtc; > > > + } > > > + > > > + /* If no CRTCs updated, then nothing to do: */ > > > + if (!wait_crtc) > > > + return; > > > > Is there an actual point in limiting this to single crtc updates? > > That immediately excludes tiled displays/etc. > > > > Handling an arbitrary number of crtcs shouldn't really be a lot > > more complicated should it? > > I guess I could find the soonest upcoming vblank of all the CRTCs and > use that as the deadline? Yeah, that seems reasonable. The flips are supposed to happen atomically (if possible) anyway so collapsing the thing to a single deadline for all makes sense to me.
diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c index d579fd8f7cb8..d8ee98ce2fc5 100644 --- a/drivers/gpu/drm/drm_atomic_helper.c +++ b/drivers/gpu/drm/drm_atomic_helper.c @@ -1511,6 +1511,40 @@ void drm_atomic_helper_commit_modeset_enables(struct drm_device *dev, } EXPORT_SYMBOL(drm_atomic_helper_commit_modeset_enables); +/* + * For atomic updates which touch just a single CRTC, calculate the time of the + * next vblank, and inform all the fences of the deadline. + */ +static void set_fence_deadline(struct drm_device *dev, + struct drm_atomic_state *state) +{ + struct drm_crtc *crtc, *wait_crtc = NULL; + struct drm_crtc_state *new_crtc_state; + struct drm_plane *plane; + struct drm_plane_state *new_plane_state; + ktime_t vbltime; + int i; + + for_each_new_crtc_in_state (state, crtc, new_crtc_state, i) { + if (wait_crtc) + return; + wait_crtc = crtc; + } + + /* If no CRTCs updated, then nothing to do: */ + if (!wait_crtc) + return; + + if (drm_crtc_next_vblank_start(wait_crtc, &vbltime)) + return; + + for_each_new_plane_in_state (state, plane, new_plane_state, i) { + if (!new_plane_state->fence) + continue; + dma_fence_set_deadline(new_plane_state->fence, vbltime); + } +} + /** * drm_atomic_helper_wait_for_fences - wait for fences stashed in plane state * @dev: DRM device @@ -1540,6 +1574,8 @@ int drm_atomic_helper_wait_for_fences(struct drm_device *dev, struct drm_plane_state *new_plane_state; int i, ret; + set_fence_deadline(dev, state); + for_each_new_plane_in_state(state, plane, new_plane_state, i) { if (!new_plane_state->fence) continue;